SCMR 22nd Annual Scientific Sessions
T1 and T2 cardiac parametric mapping can be used to characterize several cardiomyopathies including infarction and inflammation. Conventional T1 and T2 mapping are usually achieved with sequential, non-registered and inefficient 2D protocols. MR Fingerprinting (MRF)  is a novel approach that enables simultaneous, multi-parametric mapping. 2D breath-hold ECG-triggered cardiac MRF  has been shown to achieve comparable mapping quality to conventional T1 and T2 methods in reduced scan time. Here, we propose a novel 3D cardiac MRF method for improved heart coverage. Robustness to respiratory motion of 3D cardiac MRF is investigated with acquisitions performed under breath-hold and free-breathing.
With MRF, multiple sequence parameters are varied throughout the scan to sensitize the acquisition to multiple parameters (e.g. T1, T2). Tissue magnetization evolutions (fingerprints) for an MRF acquisition are simulated with Bloch equations, generating a database of spin signals (dictionary). A pixel-wise match between reconstructed MRF images and the dictionary yields the desired parameters. We propose a 3D ECG-triggered cardiac MRF acquisition using different pre-pulses (inversion recovery, no pre-pulse or T2 preparation) in different heartbeats, as depicted in Fig.1. Data is acquired with a golden angle stack-of-stars trajectory and reconstructed with a global and locally low rank MRF reconstruction  to enable high acceleration factors. Two healthy subjects were scanned with a gradient echo MRF sequence at a 1.5T scanner. Imaging parameters included: TE/TR = 1.95/4.1 ms, 2x2x6 mm3 resolution, 8 slices, linearly varying flip angle, diastolic acquisition window of ~200 ms, IR delays of 8 and 300 ms, T2prep durations of 40, 80 and 160 ms, scan time ~ 5 minutes. Acquisitions were performed under multiple breath-holds (BH), to compensate for respiratory motion, and under free-breathing (FB) to investigate the robustness of 3D cardiac MRF to respiratory motion.
T1 and T2 MRF maps with 8 slices acquired in BH and FB are shown in Figures 2 and 3, respectively. T1 and T2 are in good agreement with literature values from conventional T1 and T2 mapping approaches, with a minor underestimation for T1 (similar to the one reported before for 2D cardiac MRF). Parametric mapping was achieved in both cases, however respiratory motion artefacts (predominantly blurring) was present in the FB case.
A 3D cardiac MRF method has been proposed, showing minor underestimation in T1 which may be improved by modelling inversion efficiency. 3D cardiac MRF has also shown sensitivity to respiratory motion (FB), thus respiratory motion compensation approaches need to be considered for free-breathing acquisitions. Future work will validate the proposed approach against gold standard T1 and T2 mapping methods and investigate diaphragmatic navigator and self-gating approaches to compensate for respiratory motion.